Fire Dataset

Context

The dataset was created by my team during the NASA Space Apps Challenge in 2018, the goal was to use the dataset to develop a model that can recognize the images with fire.

Content

Data was collected to train a model to distinguish between the images that contain the fire (fire images) and regular images (non-fire images), so the whole problem was binary classification. Data is divided into 2 folders, fire images folder contains 755 outdoor-fire images some of them contains heavy smoke, the other one is non-fire images which contain 244 nature images (eg: forest, tree, grass, river, people, foggy forest, lake, animal, road, and waterfall). Hint: Data is skewed, which means the 2 classes(folders) don't have an equal number of samples, so make sure that you have a validation set with an equally sized number of images per class (eg: 40 images of both fire and non-fire classes).

Dataset

Modeling: Keras Multi-layer Perceptron (MLP) for Image Classifications

A multi-layer perceptron (MLP) is a class of feedforward artificial neural network (ANN). The algorithm at each iteration uses the Cross-Entropy Loss to measure the loss, and then the gradient and the model update is calculated. At the end of this iterative process, we would reach a better level of agreement between test and predicted sets since the error would be lower from that of the first step.

Compiling and fitting the model

Here, we only went through a few iterations; however, we need to train the model for more iterations to get more accurate results.


References

  1. TensorFlow Image
  2. Image classification
  3. Kaggle Fire Dataset
  4. TensorFlow Docs